Predictive Scheduling Laws: Where They Apply and What They Require

What a predictive scheduling law actually requires

A predictive scheduling law, often called a fair workweek law, does two things. It makes you publish the schedule a fixed number of days before the work starts, and it makes you pay a premium when you change a shift you already published.

The premium is the part that changes behaviour. Publishing two weeks out costs nothing. Moving a shift three days before it starts costs money, every time. The whole point of these laws is to stop employers treating a published schedule as a draft.

It is worth remembering what that is actually for. I spent a stretch working hourly at a large national retail chain, and I often did not know my schedule until less than two weeks out. Anything I wanted to plan around it, a class, a second job, a family commitment, came down to guessing. Almost every ordinance below arrived after that, and I would have taken any of them.

Most of them carry three more requirements, and the third is the one employers miss.

A good faith estimate of hours at hire. You tell a new employee roughly how many hours to expect and whether on-call shifts are part of the job. Get this wrong consistently and it becomes evidence in a claim.

Rest between shifts. Someone who closes cannot be scheduled to open the next morning without a minimum gap, usually nine to eleven hours, or without extra pay and, in most places, their written consent. The industry name for the pattern is a clopening. Seattle sets the gap at ten hours.

Access to hours. Before you hire anyone new, you have to offer the spare hours to the part-timers you already employ. More on this below, because it is the requirement that catches people out.

Where do predictive scheduling laws apply?

Oregon is the only state with one. Everywhere else the rules are municipal, so coverage stops at a city boundary and a chain can be covered in one location and not in the next.

Chicago is the outlier worth reading closely. It covers seven industries rather than the usual two or three, and it is the only one that also caps coverage by what the employee earns. That cap adjusts every year, so someone outside it this year can be inside it next.

Three of these arrived recently: Berkeley and Evanston in 2024, and unincorporated Los Angeles County in July 2025. They broadly follow the pattern above, but each sets its own coverage thresholds, so read the ordinance rather than assuming it matches the nearest big city.

A number of states have gone the other way and passed preemption laws that stop their cities enacting scheduling ordinances at all. If you operate in one of those, a local rule cannot appear. Preemption statutes get amended often enough that it is worth confirming rather than assuming.

How much notice do you owe?

Fourteen days is the common answer, and it is now the answer in Oregon too. The state requirement rose from seven days to fourteen on 1 July 2025.

The exception worth knowing is New York City, which runs two different regimes. Fast food employers owe fourteen days. Retail employers with twenty or more employees owe seventy-two hours, and retail is separately barred from on-call scheduling altogether.

Philadelphia is a useful reminder that these numbers move. Its ordinance started at ten days of notice in 2020 and rose to fourteen in 2021. A compliance note written against the original figure has been wrong for five years.

What does a schedule change actually cost?

Predictability pay is calculated per change, not per pay period, and the formula varies by jurisdiction. Oregon's is the cleanest to state:

• A shift that is cancelled or shortened: half the employee's regular rate for every scheduled hour they do not work.
• A start or end time that moves without reducing hours: one extra hour of pay.

Other places scale the premium by how late the change lands. San Francisco owes one hour of pay for a change made with less than seven days of notice, and two to four hours when the change comes inside 24 hours, depending on how long the shift was.

The arithmetic gets uncomfortable quickly. Take a single store under a rule like San Francisco's, paying $20 an hour, where a manager moves three shifts a week inside the 24 hour window and those shifts run longer than four hours. That is four hours of premium per change, so $80 a change, $240 a week, a little over $12,000 a year. For one location, from a habit nobody has written down as a policy.

Two things follow that matter more than the exact figures.

Employee consent usually removes the premium. Someone who asks to swap out of a shift, or volunteers to pick one up, does not generate predictability pay. The liability attaches to employer-initiated changes. Whether your records can still tell those two apart six months later is the real question.

The premium is wages, not a fine. It is owed automatically, it shows up in a wage claim like any other unpaid wages, and enforcement agencies treat a pattern of unpaid premiums as wage theft rather than a paperwork problem.

When you do not owe predictability pay

The exemptions are narrower than most managers assume, and worth knowing precisely, because "we had no choice" is not one of them on its own.

San Francisco's list is representative. You do not owe the premium when operations cannot start or continue because of threats to employees or property, a public utility failure, an act of God, another employee's unexcused absence, another employee failing to report or being sent home, or when the employee trades the shift or requests the change themselves. Requiring overtime is also excluded.

Read that list again and notice what is not on it. Being short-staffed because someone quit last week is not an exemption. A sales forecast coming in low is not an exemption. Weather that makes the day quiet, but does not close you, is not an exemption. The carve-outs cover genuine emergencies and the employee's own choices, and very little else.

The requirement most employers miss: access to hours

Nearly every ordinance on the books says that when extra hours become available, you have to offer them to your existing part-time staff before you hire anyone new. Chicago, Philadelphia, Seattle, New York and San Francisco all carry a version of it.

It is the least discussed provision, and breaching it does not feel like anything. Nobody files a complaint about a shift they were never told about. A manager posts a job opening because that is the normal way to get more coverage, and the part-timer who wanted eight more hours a week never finds out the hours existed.

Assume ignorance rather than bad faith here. Keeping up with rules that change by city, and change again every couple of years, is genuinely hard for a business without a compliance function. Most employers who miss this one do not know the obligation exists, which is a different problem from choosing to ignore it, and a much easier one to fix.

The practical version is simple. Before you post a role, the open hours go to the people already on your books, in writing, with enough time to answer. That is a scheduling habit rather than an HR process, which is exactly why it tends to get skipped.

What this changes about how you build a schedule

The operational consequence is narrower than most compliance summaries suggest. These laws do not make scheduling harder. They make editing a published schedule expensive, which is a different problem with a different fix.

Three habits do most of the work.

1. Publish later, but publish finished. A schedule published sixteen days out and then edited four times is worse, legally and practically, than one published at fourteen days and left alone. The notice period is a floor, not a target.
2. Know the cost before you click. The moment that creates the liability is a manager moving a shift inside the window. That is the moment to surface what it will cost, not the end of the pay period when it turns up on a report.
3. Record who asked. Employee-initiated changes are exempt nearly everywhere. If a swap arrives by text message, you have no evidence of who initiated it. If it arrives as a request somebody approved, you do.

This is the reasoning behind how Shift Amp handles publishing. A draft week can be edited freely, publishing is an explicit step that shows exactly what is changing and who it reaches, and every later change is attributed to the person who made it.

Worth being straight about why that exists: the audit trail was built for both reasons at once. Knowing who moved a shift settles arguments inside a business that has nothing to do with compliance, and it happens to be the exact record these ordinances make valuable. The distinction between a manager's edit and an employee's request is captured either way, which is what the law turns on.

It is also why how you structure shifts in the first place matters more under these rules than without them. A rotating shift pattern that is genuinely predictable generates almost no predictability pay. One rebuilt from scratch every week generates it constantly.

What these laws do not do

They do not cap hours and they are not overtime rules. You can still schedule a long week, and the ordinary overtime calculation is unchanged by any of this. They also do not stop you requiring extra hours, which is governed separately. See mandatory overtime for that.

They do not apply to most small employers either. Every ordinance in the table sets a headcount threshold and several add a location count on top. A single-site independent restaurant is outside all of them.

The reason to understand the rules anyway is that thresholds are counted globally, not locally. A franchisee with one location in Chicago is covered if the brand has thirty locations and 250 employees in total. Plenty of employers find out they were covered only after a complaint, which is the most expensive way to learn it.